Quantitative Immunohistochemistry in the Differential Diagnosis of Lung Tumours with Hierarchical Cluster Analysis and Ranked Comparative Graphic Immunoprofile
B Bozoky, C Fernandez, M Bjornstedt. Karolinska University Hospital, Huddinge, Stockholm, Sweden
Background: Currently, classification and differential diagnosis of lung tumours is mainly based on their histological features, in absence of a comprehensive immunohistochemical characterisation. This study was aimed to assess and prove the importance of the quantitative immunohistochemical approach as a diagnostic tool in the differential diagnosis of lung tumours.
Design: 285 lung tumours including 157 adenocarcinomas (AC), 37 squamous cancers (SC), 61 neuroendocrine tumours (NT), 19 large cell cancers (LC) and 11 adenosquamous cancers (AS) were immunostained with up to 41 conventional antibodies, semiquantitatively evaluated according to the extent of immunoreactivity (scarce:<10%; partial:10-85%; diffuse:>85%), registered in spreadsheet format and imported into a relational database. This database was used for hierarchical cluster analysis and ranked comparative graphic immunoprofiles (CGIP) for the differential diagnosis. GCIP were generated by a tailored, self developed software application and statistically significant antibodies were calculated using t-test and significance analysis of microarrays (SAM).
Results: Hierarchical clustering analysis separated three main tumour types and related specific antibody clusters: AC-specific set comprising cytokeratin (ck) 7, ck18, ck19, CEA, CA125, TTF1, EMA, Muc1, Muc5ac; SC-specific set including ck5, ck17, HMWck, p63, maspin; and NT-specific set containing CD56, chromogranin A, synaptophysin. AS and LC cancers could only be separated from the main tumour types by ranked CGIP supported by significance analysis. The statistically significant antibodies were determined for the differential diagnostic alternative of: AC-SC, AC-AS, AC-LC, SC-AS, SC-LC, AS-LC and NT against all the other tumour types.
Conclusions: Comprehensive graphic immunoprofiles for the different lung tumour types and antibody clusters specific for AC, SC and NT could be provided. Different sets of antibodies proved to be statistically significant in the differential diagnosis of the five tumour types depending on the diagnostic context. Ranked CGIP can visualize these differences and can be used as differential diagnostic tool.
Category: Special Category for 2009 - Pan-genomic/Pan-proteomic approaches to Cancer
Tuesday, March 10, 2009 9:30 AM
Poster Session III # 228, Tuesday Morning